International Journal of Innovative Research in Computer and Communication Engineering
ISSN Approved Journal | Impact factor: 8.771 | ESTD: 2013 | Follows UGC CARE Journal Norms and Guidelines
| Monthly, Peer-Reviewed, Refereed, Scholarly, Multidisciplinary and Open Access Journal | High Impact Factor 8.771 (Calculated by Google Scholar and Semantic Scholar | AI-Powered Research Tool | Indexing in all Major Database & Metadata, Citation Generator | Digital Object Identifier (DOI) |
| TITLE | EcoTrack: An AI-Powered Individual Carbon Footprint Prediction, Analysis & Advisory System Using LSTM Neural Networks, OCR Bill Scanning & Generative AI |
|---|---|
| ABSTRACT | Climate change is one of the most pressing global challenges of the 21st century. Individual human activities account for over 71% of global greenhouse gas emissions, yet most individuals have no accessible means of measuring, predicting, or reducing their personal carbon footprint. This paper presents EcoTrack, a full-stack AI-powered desktop and web platform designed to address this gap. EcoTrack enables users to calculate their monthly carbon footprint through a structured lifestyle input form, predict future emissions using a Long Short-Term Memory (LSTM) neural network, extract carbon data from utility bills via Optical Character Recognition (OCR), and receive personalized sustainability recommendations through a Generative AI chatbot (EcoCoach). An Artificial Neural Network (ANN) classifier further powers the personalized recommendations engine by identifying each user’s highest-emission domain. The system incorporates a gamified social leaderboard, active environmental challenges, and a longitudinal user profile. Experimental results demonstrate an LSTM prediction RMSE of 7.1 kg CO₂e, ANN validation accuracy of approximately 88%, OCR field extraction accuracy of 86.7%, and an API response time of 0.4 seconds post-optimization. Comparative analysis confirms EcoTrack’s novelty as the first platform to combine real-time emission calculation, LSTM forecasting, OCR-driven data ingestion, ANN-based recommendations, and Generative AI advisory in a single cohesive system. |
| AUTHOR | ZAINAB SHAIKH, SARA KHAN, VISHAKHA CHAVAN, DHANASHREE SHINDE, S.D. CHAVAN Department of Artificial Intelligence & Machine Learning, AISSMS Polytechnic, Pune, Maharashtra, India |
| VOLUME | 182 |
| DOI | DOI: 10.15680/IJIRCCE. 2026.1403064 |
| pdf/64_EcoTrack An AI-Powered Individual Carbon Footprint Prediction, Analysis & Advisory System Using LSTM Neural Networks, OCR Bill Scanning & Generative AI.pdf | |
| KEYWORDS | |
| References | [1] IPCC. (2021). Climate Change 2021: The Physical Science Basis. Contribution of Working Group I to the Sixth Assessment Report. Cambridge University Press. [2] Ritchie, H., & Roser, M. (2020). CO₂ and Greenhouse Gas Emissions. Our World in Data. https://ourworldindata.org/co2-emissions [3] Dietz, T., Gardner, G. T., Gilligan, J., Stern, P. C., & Vandenbergh, M. P. (2009). Household actions can provide a behavioral wedge to rapidly reduce US carbon emissions. PNAS, 106(44), 18452–18456. [4] Wiedmann, T., & Minx, J. (2008). A definition of ‘carbon footprint’. Ecological Economics Research Trends, 1, 1–11. [5] Druckman, A., & Jackson, T. (2009). The carbon footprint of UK households 1990–2004. Ecological Economics, 68(7), 2066–2077. [6] Hochreiter, S., & Schmidhuber, J. (1997). Long short-term memory. Neural Computation, 9(8), 1735–1780. [7] Zhang, B., et al. (2019). A novel encoder-decoder model based on LSTM for air quality prediction. Expert Systems with Applications, 131, 332–344. [8] Kong, W., et al. (2019). Short-term residential load forecasting based on LSTM recurrent neural network. IEEE Transactions on Smart Grid, 10(1), 841–851. [9] Wen, L., Yuan, X., & Li, S. (2020). Forecasting CO₂ emissions in China’s commercial department through an improved LSTM model. Energy, 218, 119435. [10] Smith, R. (2007). An overview of the Tesseract OCR engine. Proceedings of ICDAR 2007, IEEE. [11] Liu, X., et al. (2021). Graph convolution for multimodal information extraction from visually rich documents. NAACL-HLT. [12] Brown, T. B., et al. (2020). Language models are few-shot learners. Advances in Neural Information Processing Systems (NeurIPS), 33. |
